DESCRIPTION Abstract: This proposal is to pursue a highly innovative general theory that integrates evolution, developmental biology, epigenetics, and disease. Epigenetics is the study of information heritable during cell division other than the DNA sequence, and it underlies normal development and is important in cancer and possibly other common disease. I have been involved in this field since my discovery of altered DNA methylation in cancer in 1983. Since that time, I and others have grappled with how to incorporate epigenetics into evolutionary thinking. Some have proposed a Lamarckian inheritance of epigenetic marks, that is, environmentally directed epigenetic changes, which in my view is with rare exceptions an implausible mechanism over the long term because of powerful gametic reprogramming. I suggest a new inherited stochastic variation model in which genetic variants that do not result in mean observable phenotypes, but could change the variability of the observed phenotypes through epigenetic mechanisms. These heritable genetic variants for increased epigenetic variance would be found at genes in which the direction of environmental selection (positive or negative) fluctuates over long periods of time. Such variants for variation would be mediated epigenetically, for example, stochastic variation in DNA methylation in an isogenic background. At the same time, by increasing the tails at both ends of a phenotype distribution curve, there would be increased disease susceptibility in a recently changed environment, such as the Western diet. I have found three lines of preliminary data supporting this new idea: highly variable regions of DNA methylation in a given tissue in inbred mice raised in the same environment, and regulating key genes for development; mouse/human differences in DNA sequence in which one species has this variation and another does not; and a link between some these variably methylated regions and body mass index.